8307106: Allow concurrent GCs to walk CLDG without ClassLoaderDataGraph_lock #13718

fisk · 2023-04-28T13:15:20Z

A concurrent GC with concurrent class unloading can't currently walk the CLDG without the CLDG_lock today. We should add some synchronization code so it can do that safely. This patch adds the missing bits.

Progress

Change must be properly reviewed (1 review required, with at least 1 Reviewer)
Change must not contain extraneous whitespace
Commit message must refer to an issue

Issue

JDK-8307106: Allow concurrent GCs to walk CLDG without ClassLoaderDataGraph_lock

Reviewers

Stefan Karlsson (@stefank - Reviewer)
Axel Boldt-Christmas (@xmas92 - Committer)
Coleen Phillimore (@coleenp - Reviewer)
David Holmes (@dholmes-ora - Reviewer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk.git pull/13718/head:pull/13718
$ git checkout pull/13718

Update a local copy of the PR:
$ git checkout pull/13718
$ git pull https://git.openjdk.org/jdk.git pull/13718/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 13718

View PR using the GUI difftool:
$ git pr show -t 13718

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/13718.diff

Webrev

Link to Webrev Comment

…ph_lock

bridgekeeper · 2023-04-28T13:16:42Z

👋 Welcome back eosterlund! A progress list of the required criteria for merging this PR into master will be added to the body of your pull request. There are additional pull request commands available for use with this pull request.

openjdk · 2023-04-28T13:18:52Z

@fisk The following label will be automatically applied to this pull request:

hotspot-runtime

When this pull request is ready to be reviewed, an "RFR" email will be sent to the corresponding mailing list. If you would like to change these labels, use the /label pull request command.

openjdk · 2023-04-28T13:22:47Z

@fisk This change now passes all automated pre-integration checks.

ℹ️ This project also has non-automated pre-integration requirements. Please see the file CONTRIBUTING.md for details.

After integration, the commit message for the final commit will be:

8307106: Allow concurrent GCs to walk CLDG without ClassLoaderDataGraph_lock

Reviewed-by: stefank, aboldtch, coleenp, dholmes

You can use pull request commands such as /summary, /contributor and /issue to adjust it as needed.

At the time when this comment was updated there had been 57 new commits pushed to the master branch:

c8f3756: 8306729: Add nominal descriptors of modules and packages to Constants API
0b5b642: 8307150: RISC-V: Remove remaining StoreLoad barrier with UseCondCardMark for Serial/Parallel GC
418a825: 8306466: Open source more AWT Drag & Drop related tests
74667e3: 8303919: Instant.ofEpochMilli says it can throw an exception that it can't
76991c8: 8282232: [Win] GetMousePositionWithPopup test fails due to wrong mouse position
05b9b58: 8302496: Runtime.exit incorrectly says it never throws an exception
8a70664: 8293117: Add atomic bitset functions
8c106b0: 8303784: no-@target annotations should be applicable to type parameter declarations
b76f320: 8307123: Fix deprecation warnings in DPrinter
a8bf2ac: 8304888: Add dedicated VMProps for linker and fallback linker
... and 47 more: https://git.openjdk.org/jdk/compare/83a98c66f1747fec3da77578b646498c4cb5637d...master

As there are no conflicts, your changes will automatically be rebased on top of these commits when integrating. If you prefer to avoid this automatic rebasing, please check the documentation for the /integrate command for further details.

➡️ To integrate this PR with the above commit message to the master branch, type /integrate in a new comment.

mlbridge · 2023-04-28T13:33:53Z

Webrevs

00: Full (fd89b947)

coleenp

Looks good to me too. Thanks for the preview, Erik.

fisk · 2023-04-28T16:32:41Z

Thanks for the reciews @coleenp @stefank and @xmas92!

dholmes-ora

I don't see any change here where code stops acquiring the CLDG lock. Is this just a preparatory change?

Using acquire/release etc is certainly necessary to be able to walk the list safely, but it is not obvious it is sufficient. I can't tell when nodes in the CLDG actually get deleted such that the GC thread doing the walking can't access a no longer existing node.

Thanks.

dholmes-ora · 2023-05-02T09:17:50Z

src/hotspot/share/classfile/classLoaderDataGraph.cpp

@@ -520,7 +527,8 @@ bool ClassLoaderDataGraph::do_unloading() {
      prev->set_next(data);
    } else {
      assert(dead == _head, "sanity check");
-      _head = data;
+      // The GC might be walking this concurrently
+      Atomic::store(&_head, data);


If you are using load_acquire then surely this must be a release-store as they need to be paired.

The load_acquire synchronizes with release_store on insertions. This relaxed unlinking store, doesn't publish anything new that the load_acquire is interested in. It rolls back the head to a CLD that was installed at some earlier point with release_store, meaning it has already been published safely before.

Thanks for clarifying

fisk · 2023-05-02T12:47:49Z

Thanks for having a look, @dholmes-ora.

I don't see any change here where code stops acquiring the CLDG lock. Is this just a preparatory change?

Yes, indeed. It's for generational ZGC.

Using acquire/release etc is certainly necessary to be able to walk the list safely, but it is not obvious it is sufficient. I can't tell when nodes in the CLDG actually get deleted such that the GC thread doing the walking can't access a no longer existing node.

We unlink all unloading classes during do_unloading(). Then the GC performs a thread-local handshake with all threads, before it finally calls purge() to delete them. So in that sense, we use safe memory reclamation to ensure that nobody is poking around at the CLD concurrently any longer, when we delete them.
Note that we already perform release_store on the head when CLDs are inserted (always prepended). And that's because we need to safely publish the next pointer (and generally the initial CLD contents). However, when we unlink CLDs, we only use relaxed stores, because we are not mutating the CLD in any interesting way that needs ordering w.r.t. the unlinking stores.

dholmes-ora

Thanks for the additional explanations @fisk.

fisk · 2023-05-03T08:14:49Z

Thanks for the additional explanations @fisk.

Thanks @dholmes-ora!

fisk · 2023-05-03T08:14:58Z

/integrate

openjdk · 2023-05-03T08:16:24Z

Going to push as commit 462b1df.
Since your change was applied there have been 57 commits pushed to the master branch:

c8f3756: 8306729: Add nominal descriptors of modules and packages to Constants API
0b5b642: 8307150: RISC-V: Remove remaining StoreLoad barrier with UseCondCardMark for Serial/Parallel GC
418a825: 8306466: Open source more AWT Drag & Drop related tests
74667e3: 8303919: Instant.ofEpochMilli says it can throw an exception that it can't
76991c8: 8282232: [Win] GetMousePositionWithPopup test fails due to wrong mouse position
05b9b58: 8302496: Runtime.exit incorrectly says it never throws an exception
8a70664: 8293117: Add atomic bitset functions
8c106b0: 8303784: no-@target annotations should be applicable to type parameter declarations
b76f320: 8307123: Fix deprecation warnings in DPrinter
a8bf2ac: 8304888: Add dedicated VMProps for linker and fallback linker
... and 47 more: https://git.openjdk.org/jdk/compare/83a98c66f1747fec3da77578b646498c4cb5637d...master

Your commit was automatically rebased without conflicts.

openjdk · 2023-05-03T08:16:38Z

@fisk Pushed as commit 462b1df.

💡 You may see a message that your pull request was closed with unmerged commits. This can be safely ignored.

8307106: Allow concurrent GCs to walk CLDG without ClassLoaderDataGra…

fd89b94

…ph_lock

openjdk bot added the rfr Pull request is ready for review label Apr 28, 2023

openjdk bot added the hotspot-runtime hotspot-runtime-dev@openjdk.org label Apr 28, 2023

stefank approved these changes Apr 28, 2023

View reviewed changes

xmas92 approved these changes Apr 28, 2023

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label Apr 28, 2023

coleenp approved these changes Apr 28, 2023

View reviewed changes

dholmes-ora reviewed May 2, 2023

View reviewed changes

dholmes-ora approved these changes May 3, 2023

View reviewed changes

openjdk bot added the integrated Pull request has been integrated label May 3, 2023

openjdk bot closed this May 3, 2023

openjdk bot removed ready Pull request is ready to be integrated rfr Pull request is ready for review labels May 3, 2023

8307106: Allow concurrent GCs to walk CLDG without ClassLoaderDataGraph_lock #13718

8307106: Allow concurrent GCs to walk CLDG without ClassLoaderDataGraph_lock #13718

Uh oh!

Conversation

fisk commented Apr 28, 2023 • edited by openjdk bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Progress

Issue

Reviewers

Reviewing

Webrev

Uh oh!

bridgekeeper bot commented Apr 28, 2023

Uh oh!

openjdk bot commented Apr 28, 2023

Uh oh!

openjdk bot commented Apr 28, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

mlbridge bot commented Apr 28, 2023

Webrevs

Uh oh!

coleenp left a comment

Choose a reason for hiding this comment

Uh oh!

fisk commented Apr 28, 2023

Uh oh!

dholmes-ora left a comment

Choose a reason for hiding this comment

Uh oh!

dholmes-ora May 2, 2023

Choose a reason for hiding this comment

Uh oh!

fisk May 2, 2023

Choose a reason for hiding this comment

Uh oh!

dholmes-ora May 3, 2023

Choose a reason for hiding this comment

Uh oh!

fisk commented May 2, 2023

Uh oh!

dholmes-ora left a comment

Choose a reason for hiding this comment

Uh oh!

fisk commented May 3, 2023

Uh oh!

fisk commented May 3, 2023

Uh oh!

openjdk bot commented May 3, 2023

Uh oh!

openjdk bot commented May 3, 2023

Uh oh!

Uh oh!

fisk commented Apr 28, 2023 •

edited by openjdk bot

Loading

openjdk bot commented Apr 28, 2023 •

edited

Loading